Apr 16, 1996 - fibrosis, but not with any of the necro- inflammatory lesions (portal inflamma- tion, piecemeal necrosis, and lobular necrosis). TGF1l mRNA ...
98 Clin Pathol 1996;49:998-1004 998
Histological features predictive of liver fibrosis in chronic hepatitis C infection V Paradis, P Mathurin, A Laurent, F Charlotte, M Vidaud, T Poynard, C Hoang, P Opolon, P Bedossa
Abstract Aims-To
Service d'Anatomie Pathologique, Hopital de Bicetre, Paris, France V Paradis P Bedossa Service d'Hepatogastroenterologie, groupe hospitalier
Pitie-Salpetriiere, Paris, France P Mathurin T Poynard P Opolon Service d'Anatomie Pathologique F Charlotte C Hoang
which pathological features are associated with a sensitive marker of liver fibrogenesis and thus of the potential development of fibrosis in hepatitis C. Methods-The degree of liver fibrogenesis was evaluated by quantification of type I collagen mRNA and transforming growth factor (TGF) P1 mRNA (a major profibrogenic cytokine) in liver biopsy specimens from 28 patients with chronic hepatitis C and five controls, using a quantitative reverse transcription polymerase chain reaction (RT-PCR) assay. Results of mRNA quantification were correlated with histological lesions scored semiquantitatively in the same specimens. Results-Type I collagen mRNA was more strongly expressed in patients than in controls and correlated with the degree of fibrosis, but not with any of the necroinflammatory lesions (portal inflammation, piecemeal necrosis, and lobular necrosis). TGF1l mRNA concentration was higher in patients than in controls and correlated with histological grade ofactivity and lobular necrosis. This result was confirmed by in situ hybridisation experiments which showed that TGFf1 mRNA was mainly expressed in areas of focal lobular necrosis in chronic hepatitis C. Conclusion-This study shows that fibrosis, rather than necro-inflammatory lesions or activity scores, is associated with fibrogenesis and thus with potential aggravation of the fibrous deposit in chronic hepatitis C. Lobular necrosis is an important predictor of prognosis in chronic hepatitis C, as shown by its strong association with TGFP1 mRNA expression. assess
(7 Clin Pathol 1996;49:998- 1004)
CNRS URA 1484 Faculte de Pharmacie, Paris VII, France M Vidaud V Paradis P Bedossa P Mathurin T Poynard A Laurent Correspondence
to:
Dr P Bedossa, Service d'Anatomie Pathologique, H6pital de Bicetre, 78 rue du general Leclerc, 94275 Le
Kremlin-Bicetre Cedex, Paris, France. Accepted for publication 16 April 1996
Keywords: chronic hepatitis C, liver fibrogenesis, in situ hybridisation, TGFO1 mRNA.
The course of chronic hepatitis C virus (HCV) infection, even after antiviral treatment, is difficult to predict. Although the disease is frequently asymptomatic, the development of liver fibrosis is common, and cirrhosis, occurring in about one in five cases, is the major complication of this chronic disease.' There are no reliable biological or clinical indicators that can predict which patients will develop fibrosis, but if found, these would also help select appropriate patients for antiviral treatment.
Liver biopsy is used to assess the extent of liver damage, as a means of monitoring disease progression, and as a source of information for correct management. Histological examination of the liver biopsy specimens allows activity and fibrosis to be assessed.2 3 Activity gives an overall picture of the degree of necro-inflammatory lesions (grade), while fibrosis reflects the disruption to the liver architecture (stage).2 It has been suggested that in hepatitis B virus infection fibrosis is the result of activity, and thus histological activity might predict the subsequent development of fibrosis or cirrhosis.4 This hypothesis has been extended to chronic hepatitis C so that the presence of histological activity is a major criterion for starting antiviral treatment. However, this has been challenged, as data from several series of liver biopsy specimens from patients with chronic hepatitis C showed that activity is usually mild, but that fibrosis and cirrhosis are common findings.'68 As far as we are aware no studies have assessed the predictive value of these different histological features in terms of the aggravation of liver fibrosis in chronic hepatitis C. Liver tissue can also be used to investigate tissue markers of ongoing fibrogenesis, a complex process that examines the different extracellular matrix (ECM) components, such as collagens, fibronectin, and proteoglycans.' 10 In liver fibrosis and cirrhosis, collagens, especially type I collagen, are much in evidence." Furthermore, type I collagen mRNA seems to be a definitive and reliable marker for fibrogenic activity and so of potential fibrous deposition.'2 13 The aim of this study was to determine the pathological features associated with the potential progression to fibrosis in chronic hepatitis C. We therefore evaluated the degree of fibrogenesis in liver biopsy specimens by quantification of liver type I collagen mRNA, using a sensitive quantitative reverse transcriptionpolymerase chain reaction (RT-PCR) assay. We also performed semiquantitative scoring of the different pathological features present in the same liver biopsy specimens and assessed the correlations between histological scores and type I collagen mRNA concentration. As transforming growth factor (TGF) f31 has a major role in the pathogenesis of fibrosis in chronic hepatitis and cirrhosis-mainly by stimulation of fat storing cells, the main cellular source of matrix proteins'415 we also quantified its mRNA concentration in the same biopsy specimens and correlated it with histological variables. TGFP1 mRNA expres-
Features predictive of liverfibrosis in chronic hepatitis C infection
sion was also studied using in situ hybridisation and the results compared with quantitative data. Methods Twenty eight patients (nine women and 19 men; mean age 48 years, range 29-69 years) were included in the study. All patients were positive for antibody to HCV by second generation tests. Serum viral load was quantified by branched DNA (Amplec Chiron, Chriron, Emeryville, California, USA). Genotypes were also determined by a competitive oligonucelotide priming (COP) PCR assay in 21 patients.16 Serum aminotransferases (ALT), bilirubin, and y-glutamyltranspeptidase were measured on entry to the study. None of the patients had received antiviral treatment. The delay between presumed infection and time of inclusion was determined for 19 patients. Selection of patients was based on consecutive recruitment and on the availability of enough liver tissue to perform both an accurate histological study and RNA extraction for mRNA quantification. Twelve patients received interferon a 2b (Intron-A, Schering Plough, UK) in a dose of 3 mU, three times weekly for six months. Two patients were non-responders (no improvement in ALT activity during treatment), four were partial responders (decrease of less than 50% of initial value of ALT), and six were complete responders (return to normal of ALT values). Five patients without HCV infection, normal liver function tests, and neither chronic hepatitis nor fibrosis on the biopsy specimen, constituted the control group. Liver biopsy specimens were obtained for each patient using a percutaneous procedure. About half of each specimen was immediately frozen in liquid nitrogen and stored at -80°C until RNA extraction was performed; the rest of the specimen was fixed, embedded in paraffin wax and processed for histological study according to standard procedures. A set of representative paraffin wax embedded liver biopsy specimens showing chronic HCV infection with differing degrees of activity were selected for in situ hybridisation. These were cut into 5 jm thick sections. HISTOLOGICAL ANALYSIS OF LIVER BIOPSY SPECIMENS
Liver biopsy specimens were examined simultaneously by two different pathologists, neither of whom was aware of clinical and biological data except for positivity for antibody to HCV. For each liver biopsy specimen, the semiquantitative assessment of elementary features potentially present in HCV hepatitis was performed using a standardised questionnaire.'7 The following lesions were recorded: portal fibrosis (FO = no fibrosis, Fl = portal fibrosis without septa, F2 = portal fibrosis with rare septa, F3 = numerous septa without cirrhosis, F4 = cirrhosis); focal lobular necrosis (0 = less than one necro-inflammatory focus per lobule, 1 = at least one necro-inflammatory focus per lobule, 2 = several necro-inflammatory foci per lobule or
999
confluent or bridging necrosis); portal inflammation (0 = absent, 1 = presence of mononuclear aggregates in some portal tracts, 2 = mononuclear aggregates in all portal tracts, 3 = large and dense mononuclear aggregates in all portal tracts); and piecemeal necrosis (0 = absent, 1 = focal alteration of periportal plate in some portal tracts, 2 = diffuse alteration of periportal plate in some portal tracts or focal lesions around all portal tracts, 3 = diffuse alteration of periportal plate in all portal tracts). Fatty changes (micro and/or macrovesicular) were graded as 0 = absent, 1 = mild, 2 = moderate, 3 = severe. Bile duct damage was also reported. The activity of chronic hepatitis was recorded as follows: AO = no histological activity, Al = minimal activity, A2 = moderate activity, A3 = severe activity. The degree of disease activity was defined by the integrated assessment of all necro-inflammatory lesions according to an algorithm, including piecemeal necrosis as a major decision criterion and lobular necrosis as a secondary criterion. The reproducibility of this activity grading was reported as acceptable according to a previous study.'7 Finally, the histological activity index of Knodell et al'8 (periportal ± bridging necrosis, lobular necrosis, inflammation and portal fibrosis scores) was separately recorded. Total RNA was isolated from frozen liver tissue using a method derived from the procedure of Chomczynski and Sacchi.'9 The concentration and purity of total RNA were evaluated by spectrophotometry at 260 and 280 nm. The quality of RNA was also controlled by electrophoresis on a 1 % agarose gel by checking the 18S and 28S RNA bands. Type I collagen and TGFp1 mRNA of each liver biopsy specimen were quantified by a quantitative RT-PCR assay.20 Expression of an endogenously expressed RNA (transcription factor TFIID) was used as the internal control for each specimen. Each sample was normalised on the basis of its internal control content (TFIID). Reverse transcription of total RNA was performed in a final volume of 20 jtl containing lx RT-PCR buffer (1 mM each dNTP, 5 mM MgCl2, 50 mM KC1, 10 mM TRIS-HCI, pH 8.3), 1 unit of RNase inhibitor, 50 units of Moloney murine leukaemia virus reverse transcriptase (RT) (Perkin-Elmer, Foster City, California, USA), 2.5 mM random hexamers, and 1 jig of total RNA. Samples were incubated at 20°C for 10 minutes, 42°C for 30 minutes, and reverse transcriptase was inactivated by heating at 99°C for five minutes and cooled at 5°C for five minutes. Thereafter, the 20 jl volume was brought to 100 ,ul with lx RT-PCR buffer. Quantitative PCR is based on coamplification of specific cDNA with a quantitative DNA standard (QDS) using the same pair of primers.20 Specific QDS were generated by creating a 12 base pair insertion in the wild type cDNA sequence for each target gene. QDS were constructed by PCR from cDNA using a composite primer. With this procedure, target cDNA and QDS yield labelled PCR products of different sizes which were identi-
Paradis, Mathurin, Laurent, Charlotte, Vidaud, Poynard, et al
1000
Table 1 Oligonucleotide primer sequences and PCR product sizes (base pairs) Gene
Primer sequence
PCR product size
Type I collagen
5'-TCCCCAGCCACAAAGAGTCTACA-3' 5'-F-GTGATTGGTGGGATGTCTTCGTC-3' 5'-CAACAATTCCTGGCGATACCTCA-3' 5'-F-GGTAGTGAACCCGTTGATGTCCA-3' 5'-F-ACAGGAGCCAAGAGTGAAGAA'3' 5'-CCAGAAACAAAAATAAGGAGA-3'
155
TGF,1 TFIID
199 260
F = Fluorescein molecule.
fled by a DNA sequencer. We ensured that the amplification efficiency between target cDNA and QDS was maintained during the exponential and non-exponential phases. Selection of PCR primers was performed using the OLIGO program from Bioscience Inc. (Plymouth, Minnesota, USA). Fluorescein labelling was carried out during primer synthesis by adding Fluorprime (Pharmacia, Uppsala, Sweden). The sequence of each primer and PCR product sizes are presented in table 1. For each PCR run, a master mix was prepared on ice using lx PCR buffer (2 mM MgCl2, 50 mM KC1, 10 mM TRIS-HCI, pH 8.3), 10 pM each dNTP, 200 gM each primer, 1.25 units of Taq polymerase (Perkin Elmer) and a known amount of the corresponding QDS. The QDS amount for each target cDNA was chosen to obtain similar fluorescence intensities for standard and corresponding cDNA derived PCR products. Ten microlitres of diluted RT were added to 40 gl of the PCR master mix. Amplification was achieved by 30 cycles for type I collagen and TGF,B1 and 35 cycles for TFIID (94°C, 30 seconds; 55°C, 30 seconds; 72°C, one minute), followed by a final incubation of five minutes at 72°C. In each PCR run a control without template was run in parallel. All PCR reactions were performed using a Perkin Elmer 9600 thermocycler. Fluorescence PCR products were coelectrophoresed with internal size standards (Genescan 2500 Rox, Applied Biosystems, Foster City, California, USA) through a 6% denaturing gel run on a 373A DNA sequencer (Applied Biosystems) and analysed using Genescan 672 Software (Applied Biosystems). PCR products of target cDNA and corresponding QDS are represented by two different peaks of fluorescence. The area of each peak, expressed in relative fluorescence units (RFU), was correlated with the amount of PCR product. Results for each gene were expressed as a ratio between target cDNA RFU and its corresponding QDS RFU. Experiments were performed twice with three sets for each data point. Results are presented as mean (SD) unless otherwise stated. Variability of the different assessments was estimated by standard deviation and coefficient of variation. Statistical analysis was performed using Student's t test and the Mann-Whitney U test. When necessary, statistics were performed after logarithmic transformation of the data. Correlations between type I collagen, TGF 1 mRNA concentrations with the different histological scores, clinical and biological data were performed using Pearson's test.
IN SITU HYBRIDISATION
A cocktail of single stranded oligonucleotides, chemically synthesised, and modified at the 5' end with digoxigenin by the manufacturer (R and D Systems Inc., Minnesota, USA) was used. It consisted of an equimolar mixture of three exon specific probes based on the antisense sequence of exon 6 (28 mer), exon 7A (27 mer), and exon 7B (26 mer) of the TGFl1 gene. In situ hybridisation was performed on paraffin wax embedded liver sections, as described before.2' Briefly, sections were dewaxed in xylene and graded ethanol, rinsed in DEPC water for five minutes, and treated with proteinase K at a final concentration of 10 ,ug/ml for 10 minutes at 37°C in 50 mM TRIS, pH 7.6. Slides were post-fixed in 0.4% paraformaldehyde in lx PBS for 20 minutes at 4°C. Liver sections were then hybridised overnight at 370C in 4x SSC, 50% formamide, lx Denhart's solution, 5% dextran sulphate, 0.5 mg/ml salmon sperm DNA, 0.25 mg/ml yeast tRNA containing 500 ng/ml of digoxigenin labelled TGFP1 antisense probe. After washing in decreasing concentrations of SSC (sodium citrate/sodium chloride) (twice for five minutes in 4x SSC at 37°C, twice for five minutes in 2x SSC at 37°C, and once in lx SSC for 15 minutes at room temperature), tissue sections were incubated with anti-digoxigenin antibody at a 1 in 500 dilution (Boehringer Mannheim, Mannheim, Germany) in 100 mM TRIS-HCI, 150 mM NaCl, pH 7.5, for 20 minutes at 37°C. After two rinses in the same buffer for five minutes and washing in DEPC-water with levamisole 1 mM, alkaline phosphatase was detected using 5-bromo-4-chloro-3-indolyl phosphate and nitro blue tetrazolium chloride. Digoxigenin labelled sense oligonucleotide as well as sections incubated without probe were used as negative controls. Liver sections incubated with a digoxigenin labelled poly-dT oligoprobe and recombinant CHO cell lines expressing TGF31 incubated with the TGFP1 antisense probe were used as positive controls. These controls were systematically processed in each experiment and under the same conditions.
Results The mean delay between presumed infection and liver biopsy was 16.8 months (range seven to 40 months) in the 19 patients for whom data were available. The mean (SD) values of serum ALT, bilirubin, and y-glutamyltranspeptidase were, respectively: 2.9N (2.1), 9.4 (3.61) ,umol/l and 5N (0.8) (N = upper limit). Ten patients were genotype lb, three were genotype la, one was genotype 2a, five were genotype 3a, and the other two could not be determined using the current procedure. For each liver biopsy specimen, the principal histological features and the degree of activity according to the METAVIR and Knodell index are reported in table 2. According to METAVIR, fibrosis was graded as Fl in 10 patients, F2 in nine patients, F3 in eight, and F4 in one.'7 One patient had no activity (AO), 12 had
Features predictive of liverfibrosis in chronic hepatitis C infection
1001
Table 2 Clinical, biological and histological data for 28 patients with HCV infection and five controls Patient number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Type I collagen mRNA *
TGFI31
Fibrosis
Sexlage
mRNA *
scoret
Activity gradef
Knodell index
F/64 M/57 M/54 M/44 F/69 M/37 M/61 F/64 M/68 M/68 M/36 MJ40 M/35 M/30 F/33 M/58 M/57 F/63 M/29 F/29 F/54 M/41 M/56 M/36 M/47 F/34 F/43 M/37
1.37 6.18 7.37 0.79 3.67 1.4 1.2 0.62 0 0.72 0.99 0 1.9 5.71 1.1 2.88 2.29 2.31 1.48 0.45 1.63 1.86 0.54 0.59 0.19 0.28 0.4 0.27
0.45 0.71 0.57 0.16 0.66 0.17 0.35 0.94 0.89 0 0.64 0.25 0.36 0.78 0.42 0.88 0.54 0.76 0.29 0.2 0.57 0.97 0.66 0.46 0.37 0.35 0.26 0.04
3 4 3 2 3 3 1 1 3 1 1 1 1 1 2 2 2 1 3 1 3 3 1 2 2 2 2 2
2 2 2 2 1 1 2 2 3 0 2 1 1 1 1 2 1 2 2 1 1 2 2 1 1 1 2 2
10 10 11 10 8 5 6 10 11 5 6 8 6 6 7 12 5 8 10 5 7 10 10 7 8 10 10 10
M/49 M/33 M/2 F/43 M151
0.51 0.5 0.34 0.45 0.36
0.16 0.13 0.28 0.1 0.07
0 0 0 0 0
0 0 0 0 0
0 0 0 0 0
Controls 29 30 31 32 33
*Results of mRNA quantification are normalised to TFIID mRNA content in the same sample. tO = No fibrosis; 1 = portal fibrosis without septa; 2 = portal fibrosis with rare septa; 3 = numerous septa without cirrhosis; 4 = cirrhosis. tO = No histological activity; 1 = minimal activity; 2 = moderate activity; 3 = severe activity.
Table 3 Correlations between type I collagen and TGFf3J mRNA and histologicalfeatures in liver biopsy specimens ofpatients with chronic hepatitis C Type I collegen mRNA
TGFI1 mRNA
Histology
r*
p value
r*
p value
Fibrosis score Lobular necrosis Piecemeal necrosis Portal inflammation Activity (grade) Knodell index
0.39 0.21 0.02 0.10 0.13 0.18
0.03 NS NS NS NS NS
0.29 0.47 0.27 0.20 0.53 0.43
0.01 NS NS 0.003 0.02
0.08
*Pearson's correlation coefficient.
Aft ~
~
.^.
~
~
~
Wo
Figure 1 In situ hybridisation of TGFIl in liver biopsy specimens of chronic active hepatitis C. TGFf3J mRNA was detected in some inflammatory and sinusoidal cells of the periportal area (arrow). PT = portal tract.
mild activity (Al), 14 had moderate activity (A2), and one had severe activity (A3). The mean value of the Knodell index was 8.3 (2.2) (range 5-12).' Type I collagen and TGFf1 mRNA were quantified in the livers of 28 patients with chronic hepatitis C and in five controls. The concentrations of type I collagen and TGFI1 mRNA in each subject were normalised to that of the TFIID mRNA value in the same biopsy sample. The mean coefficients of variation of triplicate assessments of TFIID, type I collagen, and TGF3l mRNA in the 30 samples were, respectively: 0.06 (range 0.02-0.10), 0.1 1 (0.04-0.15), and 0.06 (range 0.01-0.09). In the control group the mean concentration of type I collagen mRNA was 0.43 (0.08) and that of TGF 1 mRNA was 0.15 (0.08). In patients with chronic hepatitis C type I collagen mRNA was 1.7 (1.9), which was significantly higher than that in the control group (p < 0.01). The mean TGFP1 mRNA in patients with chronic hepatitis C was 0.5 (0.3), significantly higher than that of the control group (p < 0.01). The concentration of TGFP1 mRNA correlated significantly with type I collagen mRNA concentration (r = 0.45; p = 0.02). CORRELATIONS BETWEEN TYPE I COLLAGEN MRNA AND BIOLOGICAL AND HISTOLOGICAL DATA
In chronic hepatitis C patients, the concentration of type I collagen mRNA correlated with neither the biological data nor the delay in contamination. No correlation was observed between type I collagen mRNA and the HCV genotype (1.8 (2.1) and 1.2 (1.1) in the lb compared with the non-lb genotype groups). Nor was there any correlation observed with the extent of HCV viraemia. Among the different primary pathological features assessed, fibrosis was the only lesion to be correlated with the type I collagen mRNA concentration (r = 0.39; p = 0.03). The mean concentration of type I collagen mRNA was significantly lower in the 20 patients with Fl or F2 (1.2 (1.3)) than in the eight patients with F3 or F4 (2.9 (2.6); p = 0.03) fibrosis. No correlation was observed with other histological features. Results are shown in details in table 3. Among patients with significant fibrosis (F>F2), mean type I collagen mRNA was not significantly different between patients with no or mild activity (AO or Al, n = 8, 1.4 (1.2)) and those with moderate or severe activity (A2 or A3, n = 10, 2.3 (2.5)). To analyse in more detail the correlation between type I collagen mRNA and histological features, the patients were divided into two groups according to the concentration of type I collagen mRNA: those with higher than the median value (>1.1, n = 15) and those with lower than the median value (0.4, n = 16) and those with below (
